series of municipal-level vulnerability studies that draw from high-resolution environmental datasets that can capture the scale of the household. In this way, households need not be sampled from a given municipality, and a more representative picture can emerge of both household- and municipal-level processes.
Of course, remote sensing data represent only a subset of the potentially necessary data for the study of coupled human–environment system vulnerability and resilience. Other means of data collection and analysis (e.g., interviews, surveys, focus groups, participant observation) have proved valuable as well. Such methods can be effective for making observations about exposures, sensitivities, and adaptive capacities, at multiple scales. Indeed, the data generated by such approaches are often rich in information about cross-scale interactions, and can shed light on how the outcomes that are observable at one scale are associated with factors at other scales. Developing methods for assessing these cross-scale interactions would advance our understanding of the challenges faced by coupled human–environment systems. Building on past vulnerability studies that use a variety of data collection and analysis methods, and on geographical technologies that identify or estimate the variable impacts of multiple processes in individual places, the geographical sciences are well positioned to test the long-held geographical hypothesis that “scale matters” in the vulnerability domain.
Vulnerability and resilience assessments are wide-ranging. Some studies have examined agricultural dynamics in developing countries, whereas others have focused on urbanization challenges in Western Europe. The topical diversity of studies makes it difficult to draw comparisons and make generalizations from independent study results. To move forward, research is needed that can facilitate efforts to specify the general conditions under which coupled human–environment systems become vulnerable to the effects of human–environmental changes, document and assess the geographical patterns of these vulnerabilities, and analyze why such patterns emerge. If this challenge is not taken up, the vulnerability concept could evolve into an appealing idea with limited applied scientific value beyond providing spatially and temporally contingent findings from a set of noncomparable case studies.
It follows that a major research need for the coming decade is to inventory the existing vulnerability and resilience literature, and to determine whether the studies can be compared using a meta-analytic framework for the purposes of drawing broad conclusions about vulnerability—even when the data and the methods used to collect and analyze the data differ. There is a well-developed meta-analysis literature in the social sciences, specifying how to pool results from independent studies into a larger dataset that permits more powerful inferences and generalizations. Meta-analysis is most helpful and easiest when the predictor and outcome variables are similarly defined and measured. For example, medical studies of the effects of smoking on human health can be readily pooled because a person’s smoking behavior and health are variables that lend themselves well to common definitions and measurements.
Extending classical meta-analytic techniques would be useful in the case of vulnerability, where the definitions and measurements of variables are often not comparable across studies. Geographical scientists have contributed to an allied effort in recent years, attempting to draw systematic comparisons across studies of tropical deforestation that look at different variables and employ diverse methods. Notable examples of such meta-analyses include Geist and Lambin (2002), Misselhorn (2005), and Rudel (2005). None of these applications, which use techniques such as qualitative comparative analysis (QCA; Ragin, 1987), address vulnerability as such, but they hold promise for vulnerability research. Research is needed to demonstrate whether a creative blending of techniques and methods (e.g., coupling QCA with the vulnerability scoping diagram [Polsky et al., 2007]) can facilitate the production of vulnerability meta-analyses. Geographical scientists are well positioned to contribute to this undertaking. They could pool results from the emerging body of knowledge on, for example, the land-use-policy/ coastal-storm vulnerability relationship in the New Orleans area and test the applicability of the common themes that have emerged in that domain to other coastal-zone metropolitan areas exposed to similar environmental hazards (e.g., Miami, Houston).